Visually identifying similarities in a data set can be cumbersome if the number of data points to consider is large. For example, images can be sorted based on one or more attributes of the images, such as a color value, and then viewed in sorted order in a continuous sequence so that users can see images with similar color values grouped together consecutively. However, for a large number of images, it can be impossible to view all of the images at once without scaling them small enough so that they fit across the width of a computer display, for example. But the images may be scaled so small that they lose any meaningful detail. Another way to present a sorted sequence of images is to break the sequence into a number of rows. But if the images are broken up by row by row, then the images above and below a current image may not have a close color value to a current image, hence it can be difficult for users to discern which images have color values close to one another.